NAFC | Fisheries and Oceans Canada | 2019-01-16

Outline

  • Plotting options
  • Grammar of graphics
  • ggplot2: the basics
  • ggplot2: intermediate
  • Summary

Plotting options

“The greatest value of a picture is when it forces us to notice what we never expected to see.” — John Tukey

  • Spreadsheets (e.g., Excel)
  • Stand alone program (e.g., Sigma Plot)
  • Other stats programs (e.g., SAS)
  • R
    • Base
    • Various functions
    • ggplot2 (part of the tidyverse)

Grammar of graphics

  • ggplot2 is based on the grammar of graphics: the idea that you can build every graph from a few components:
    • Data
      • Variables mapped to aesthetic properties (e.g. x, y, size, colour)
    • Geometry
      • Visual markers (e.g. points, lines, polygons)
    • Coordinates
      • A coordinate system (e.g. Cartesian, polar)

ggplot2: the basics

  • Data
  • Syntax
  • Scatterplot
  • Line plot
  • Boxplot
  • Stacked bar
  • Exercise

ggplot2: the basics - data

  • ggplot2 is built for tidy data
  • Below we import and tidy up lobster movement data
library(readxl)
library(dplyr)
library(tidyr)
library(ggplot2)

duck <- read_excel("data/Duck Islands Movement.xlsx")
round <- read_excel("data/Round Island Movement.xlsx")
tags <- bind_rows(duck, round) %>%   # stack both data sets
    select(SIDE:NOTCH) %>%           # select main columns
    drop_na(SIDE, ISLAND, YEAR, TAG) # drop empty rows

ggplot2: the basics - syntax

ggplot(data = tags, mapping = aes(x = YEAR, y = LENGTH)) + # Data
    geom_point() +                                         # Geometry
    coord_cartesian()                                      # Coordinates
  • Data
    • Here the ggplot function starts the plot
      • data is supplied first
      • mapping is specified second using the aes (aesthetics) function
  • Geometry
    • The function geom_point adds points to the base layer
  • Goordinates
    • Finally, coord_cartesian defines the coordinate system
      • This is the default so it is not usually specified

Note that the pipe operator (%>%) is not used here, rather you add items (+)

ggplot2: the basics - scatterplot

ggplot(tags, aes(x = YEAR, y = LENGTH)) + 
    geom_point()

ggplot2: the basics - line plot

ggplot(tags, aes(x = YEAR, y = LENGTH, group = TAG)) + 
    geom_line()

ggplot2: the basics - box-whisker plot

ggplot(tags, aes(x = SEX, y = LENGTH)) + 
    geom_boxplot()

ggplot2: the basics - stacked bargraph

ggplot(tags, aes(x = YEAR, fill = ISLAND)) +
  geom_bar()

ggplot2: the basics - exercises

ggplot(tags, aes(x = YEAR, y = LENGTH, group = TAG)) + 
    geom_line()

Add the colour aesthetic to the above code make the plot below

ggplot2: intermediate

Move away from defaults to publication quality

  • Build upon an object
  • Change size and colour
  • Facets
  • Axes
  • Legends
  • Themes

ggplot2: intermediate - object + geom

p <- ggplot(tags)
p + geom_line(aes(x = YEAR, y = LENGTH, group = TAG, colour = SEX))

ggplot2: intermediate - object + alternate aes

p + geom_line(aes(x = YEAR, y = LENGTH, group = TAG, color = ISLAND))

ggplot2: intermediate - object + alternate aes

p + geom_line(aes(x = YEAR, y = LENGTH, group = TAG, color = SIDE))

ggplot2: intermediate - object + alternate geom

p + geom_bar(aes(x = YEAR, fill = ISLAND), stat = "count")

ggplot2: intermediate - object + geom + facet

p + geom_bar(aes(x = YEAR)) +
    facet_grid(~ ISLAND) # split plots by island

ggplot2: intermediate - object + geom + facet

p + geom_bar(aes(x = YEAR)) +
    facet_grid(SEX ~ ISLAND) # split plots by sex and island

ggplot2: intermediate - object + geom + facet

p + geom_bar(aes(x = YEAR, fill = factor(MONTH))) +
    facet_grid(SEX ~ ISLAND)

ggplot2: intermediate - object + more grammar

  • Let’s move away from the defaults by iteratively modifying the plot in the previous slide
## Save the first step into an object called p1
p1 <- ggplot(tags) + 
    geom_bar(aes(x = YEAR, fill = factor(MONTH))) +
    facet_grid(SEX ~ ISLAND)

ggplot2: intermediate - object + more grammar

## Here's what the defaults look like
p1

ggplot2: intermediate - object + more grammar

p2 <- p1 + xlab("Year") + ylab("Number of records") # change x and y labels
p2

ggplot2: intermediate - object + more grammar

p3 <- p2 + scale_fill_brewer(palette = "RdBu", name = "Month")  # name legend and use greyscale
p3

ggplot2: intermediate - object + more grammar

p4 <- p3 + coord_cartesian(ylim = c(0, 700), expand = FALSE) # set y limits and remove buffer
p4

ggplot2: intermediate - object + more grammar

p5 <- p4 + theme_bw() # use the black and white theme
p5

ggplot2: intermediate - object + more grammar

p6 <- p5 + theme(panel.grid.major = element_blank()) # remove major grid lines
p6

ggplot2: intermediate - object + more grammar

p7 <- p6 + theme(panel.grid.minor = element_blank()) # remove minor grid lines
p7

ggplot2: intermediate - object + more grammar

p8 <- p7 + theme(strip.background = element_blank()) # remove facet strip background
p8

ggplot2: intermediate - object + more grammar

  • All together plus an option for changing the facet labels
## Set-up nice island and sex labels
island_names <- c("duck" = "Duck",
                  "round" = "Round")
sex_names <- c("F" = "Female",
               "M" = "Male")

## Store the plot in an object
p <- ggplot(tags) +                                                 # set-up base layer
    geom_bar(aes(x = YEAR, fill = factor(MONTH)), stat = "count") + # add bars by year and filled by month
    facet_grid(SEX ~ ISLAND,                                        # facet by sex and island
               labeller = labeller(SEX = sex_names,                 # edit facet labels
                                   ISLAND = island_names)) +        
    xlab("Year") + ylab("Number of records") +                      # change x and y labels
    scale_fill_brewer(palette = "RdBu", name = "Month") +         # name legend and use greyscale
    coord_cartesian(ylim = c(0, 700), expand = FALSE) +             # set y limits and remove buffer
    theme_bw() +                                                    # use the black and white theme
    theme(panel.grid.major = element_blank(),                       # remove major grid lines
          panel.grid.minor = element_blank(),                       # remove minor grid lines
          strip.background = element_blank())                       # remove facet strip background

ggplot2: intermediate - object + more grammar

p

Summary

  • ggplot2 is a very flexible visualization tool
    • Have only shown a small subset of what it can do
  • Plots can help you discover patterns in your data
    • They can also uncover issues in the data
  • Your ability to explore your data will expand when you pair ggplot2 and dplyr